sam-faces

• Outcome: Review
• Latest audit: May 29, 2026 at 9:36 PM UTC
• Version:
  1.1.0

Overview

This skill openly provides local face recognition, but it automatically identifies people in photos and retains face-related data without enough consent or retention controls.

Review before installing. Use this only in workspaces where local biometric recognition is acceptable, and avoid processing photos of people who have not agreed to be identified or enrolled. Check and periodically delete the workspace face database and unknown-face crops if you do not want retained biometric data.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill explicitly instructs the agent to automatically save user-shared images to a temp path and run face identification, but it does not require any user notice or consent for biometric processing or for local storage of face-related data. Because face recognition and identity memory involve sensitive biometric information, silent processing and storage increase privacy risk and can surprise users who intended to share an image for general understanding rather than identity analysis.

Natural-Language Policy Violations

High

• Confidence: 98% confidence
• Finding: The skill directs automatic identity recognition whenever a user sends a photo, treating image sharing as implied consent for biometric identification. In context, this is more dangerous because the skill is specifically designed to build persistent identity memory and inject recognized identities into LLM context, enabling unannounced profiling, misidentification harms, and disclosure of who appears in images without explicit permission from the user or the depicted individuals.

VirusTotal

65/65 vendors flagged this skill as clean.